In the __netpoll_setup() function, when accessing the device's npinfo
pointer, replace rcu_access_pointer() with rtnl_dereference(). This
change is more appropriate, as suggested by Herbert Xu[1].
The function is called with the RTNL mutex held, and the pointer is
being dereferenced later, so, dereference earlier and just reuse the
pointer for the if/else.
The replacement ensures correct pointer access while maintaining
the existing locking and RCU semantics of the netpoll subsystem.
Link: https://lore.kernel.org/lkml/Zz1cKZYt1e7elibV@gondor.apana.org.au/ [1]
Suggested-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
---
v2 RESEND:
- Resending the same v2, since I sent it when the net-next was closed. :-P
- Link to v2: https://lore.kernel.org/r/20241122-netpoll_rcu_herbet_fix-v2-1-93a41fdbb19a@debian.org
Changes in v2:
- Targeting net-next instead of net
- Added the Acked-by and Reviewed-by
- Added the link in the commit summary
- Link to v1: https://lore.kernel.org/r/20241121-netpoll_rcu_herbet_fix-v1-1-457c5fee9dfd@debian.org
---
net/core/netpoll.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 2e459b9d88eb52b7c5e4710bf3f7740045d9314b..99e5aa9cc992f429eecf20aeadd04dc293b8f22b 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -634,7 +634,8 @@ int __netpoll_setup(struct netpoll *np, struct net_device *ndev)
goto out;
}
- if (!rcu_access_pointer(ndev->npinfo)) {
+ npinfo = rtnl_dereference(ndev->npinfo);
+ if (!npinfo) {
npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
if (!npinfo) {
err = -ENOMEM;
@@ -654,7 +655,6 @@ int __netpoll_setup(struct netpoll *np, struct net_device *ndev)
goto free_npinfo;
}
} else {
- npinfo = rtnl_dereference(ndev->npinfo);
refcount_inc(&npinfo->refcnt);
}
---
base-commit: 65ae975e97d5aab3ee9dc5ec701b12090572ed43
change-id: 20241121-netpoll_rcu_herbet_fix-3f0a433b7860
Best regards,
--
Breno Leitao <leitao@debian.org>